Knock sensor sensitivity adj.?

[mosk]

Lance, Craig, or anyone,

File this as a future feature request, but it would be great if there was some way to adjust the sensitivity of the knock sensor. Maybe this isn't possible with the GM ESC module; I don't really know, but on my noisy SOHC four banger, I find that I'm getting too many false positive hits to be able to use the knock sensor -- it's noisy enough that it retards timing at idle.

As a result, I will probably be adding a J&S SafeGuard knock sensor, which is a nice enough piece that i don't mind adding it, but I lose the ability to log knock sensor activity.

Does anyone have any thoughts on this? I know for most of you this isn't an issue, as you're using GM parts on a GM engine, but for those of us using these same parts on non-GM applications, things are tougher. Or maybe someone knows of a GM four cylinder knock sensor/ESC module combo that would work with the SP???

Jeff

 

[Lance]

The factory ESC module is the amplifier and generates a discrete pulse as the ESC signal to the ECU when it detects knock. It's just not possible to adjust it for sensitivity. GM makes several different modules with different sensitivities and that may be an option. You could also try a simple RC passive filter on the input to the ESC module (with a pot for sensitivity?). This would be pretty simple and probably do what you want. It would only reduce sensitivity however. I could post a simple schematic if you want to try it with radio shack parts.

Lance

 

[mosk]

Thanks for explaining that, Lance. Yes, if you wouldn't mind posting that schematic, I would appreciate it. If it matters, I am using an ESC module from a 87 GN.

So just to be clear, GM uses different ESC modules for different applications, and each module has different amplification characteristics? Similarly, GM uses different knock sensors, and each of these has different acoustic characteristics? I guess I didn't understand the relationship between the ESC module and the sensor. I feel a little sheepish

When I was installing my system, I spoke with Harry at PTE, and explained what I had and what I thought I needed, and he sold me an ESC module for an '87 GN. I guess I should have trusted my instincts and searched out a module for some sort of GM SOHC 4 cylinder...

Any suggestions on which engine might be a good candidate for a compatible ESC module? I'll try your schematic first, but just in case I can get that to work...

Thanks, Lance!

Jeff

 

[Lance]

Here you go...

This may be a bit technical and I really don't have good starting information to work from regarding frequency response but maybe start with a 1K pot and a 1uf - 5uf cap. The rest of you electrical type folks should chime in here.

The product RC is defined as the time constant of the charging circuit and has the symbol t. During one time constant the voltage rises to 63.2 % of its final value, while the current falls to 36.8 % of its initial value.

 

[mosk]

Thank you, Lance!!!

Seems simple enough once you draw it out I guess the truth will be in the tuning, but that's where the truth usually rests, anyway.

Thanks again,

Jeff

 

[TurboTR]

I've been giving this some thought and I hope no one takes any of this the wrong way. Here goes.

One, if the knock system is false tripping while driving along then that means there's plenty of acoustic noise energy in the engine system that's at the same frequency as the desired knock signal. The electronics have absolutely no way to tell the difference, nor can you mask the noise without also masking the real signal, which is the problem you are having. So if you go muting the sensor you may be able to quiet the false knocking but will also quiet any real signal just the same. Bottom line, noise is rendering the current knock system pretty much useless to you. And after all is said and done, if you do get knock hits at wot how will you know if it's real or not?

Next, if the system works like I think it does then what it does is take the knock sensor signal, which is just a piezo audio microphone basically, splits that signal into 2 internal channels. Each channel is filtered differently, with the intent of having channel A be an integrated or low pass adaptive background noise level reference and channel B is a bandpass channel to pass the desired knock signal, probably centered around 7-10KHz. Finally the 2 channel outputs are compared, and when there is enough energy present in the signal channel compared to the background noise channel the ESC decides that knock is occurring and signals the ecu.

Point is, if you reduce the input to the circuit you can see that you would also reduce both internal channels, and since they are compared to each other the result of trying to reduce the sensor input is questionable at best.

The sensitivity filter shown is an adjustable lowpass circuit. If the ESC works as outlined above you might also consider trying a bandstop circuit, centered at the knock frequency, or better yet at the noise frequency if you could determine it. That would certainly reduce the false hits, but again also the real signal too. That kindof defeats the reason for having it all in the 1st place. At the engine research institute we might have to spend time moving the sensor to a more favorable spot, quiet the noisy parts of the engine and also tweak the circuit parameters to maximize the signal to noise ratio. Harris Semiconductor used to offer a knock system development kit; kinda like an eval board for a microcontroller. I still have one laying around somewhere.

Acoustic knock schemes are tough to begin with, and to work well they have to be pretty specifically tuned to the platform they are on.

IME with running a solid lifter cam for the last 3 years with the Buick ESC system is that the system basically doesn't work well anymore It false hits at part throttle but never says a word at wot, and I think the thing is just overwhelmed at wot with too much noise. It can't pick any knock signal out anymore, which is kinda like trying to hear a whisper in Mile High stadium, excuse me Invesco stadium ;-) after the Broncs score a touchdown Solution- Ignore or toss it and just run high enough octane fuel when running the car hard, and always conservative timing Works for me.

Sorry for rambling and hope some of the above helps.

TurboTR

 

[mosk]

Thanks for that explanation. While I know that the sensor itself is basically a microphone, I really didn't have a good idea as to how the ESC circuit was designed to operate. Your description makes a lot of sense, and I can see the problems that might develop from simply turning down the input sensitivity. Sigh...I think I need to seek out a sensor and ESC combo that may be more appropriate for a SOHC four cylinder, one that might be pre-tuned to operate at a different frequency.

Part of my problem, if you want to call it a problem, is that I am using an engine that doesn't have a strong support community. While the basic versions of this engine, the Toyota 22R (carb) and 22RE (EFI), have been used in literally millions of vehicles, the turbocharged version (22RTE) was a much rarer beast. I can count on one hand the folks that I know that have 22RTE engines that are more radical than my own, and none of them are running a SpeedPro. So I am learning as I go along, and I often feel like I am making my own trail.

My engine has a mix of parts on it, some custom, some GM, and some Ford, so I suppose I shouldn't be too surprised when the first off-the-shelf sensor/ESC combo I try doesn't quite work out...

Anyway, I appreciate you taking the time to explain how these things work, as it shines some light into an area that I didn't understand all that well.

Jeff

 

[Lance]

I'm not really certain how the Knock Sensor and ESC module interact and TurboTR may well be right. The mere thought of a knock sensor on a high performance engine makes me shudder. Especially if good gas is used. The circuit described above will reduce the sensitivity of the high frequency noise depending on the selected cutoff frequency but that may also mean that any real detonation is squelched too. There are a lot of factors involved. Considering the high BMEP (cylinder pressure) related to a high torque engine, and using gasoline that results in a higher threshold for detonation, this is a recipe for some real damage potential. When the engine actually does detonate, the BMEP will already be very high and the resultant pressure spike from the detonation will be at a level that can cause some serious parts breakage. I hope you use good parts. Food for thought. I guess I should have stuck to my usual discourse on the subject.

With that said, it is usually informative and fun to delve into an experiment such as this. It may well do what you need it to and it is a fairly simple matter to go from a low pass to a high pass by simply swapping the 2 components around. I hope you figure it all out. Good luck.

Lance

 

[mosk]

Thanks, Lance; I do appreciate your input and help with this. I've read the online help, and I know that the official FAST position is that a knock sensor is a poor choice on a high performance engine. I even understand why a knock sensor is a poor safety net on a high performance engine, however, this being a street engine tuned for street use, I am NOT going to be running it near the ragged edges of maximum timing and maximum boost. The engine itself has been built with as much care and attention to detail as I could afford, given the power goals I am shooting for. The pistons and crank are forged, the rods have been race-prepped and hardened, and the block has been o-ringed. We built the bottom end to stand up to 450-500 hp, and I am "only" shooting for 300 rwhp, so I think I have an adequate safety margin.

Anyway, thanks again for the schematic. I'll come up with something that will work for my needs.

Jeff

 

[TurboTR]

re:

Yeah I remember a few years ago Lance had stated his opinion on the ESC on a high perf engine and I didn't quite understand how he could feel that way about it ;-)

But I feel better about it now after some experience. Plus getting to tear down and examine your turbo engine after several years of running it at high boost and all and seeing that everything was fine after all helps confirm.

TurboTR